473,387 Members | 3,801 Online
Bytes | Software Development & Data Engineering Community
Post Job

Home Posts Topics Members FAQ

Join Bytes to post your question to a community of 473,387 software developers and data experts.

DB2 ICE ( Integrated Cluster Environment )

Is there anybody out there that has implemented or will be implementing the new DB2 MPP clustering on Linux?

http://www-3.ibm.com/software/data/db2/linux/ice/

Any success stories? Disasters?

Thanks in advance,

Tim

tim-at-datad-dot-com

Nov 12 '05 #1
10 3969
"Tim Schaefer" <we*******@datad.com> wrote in message
news:E9*****************@fe09.atl2.webusenet.com.. .
Is there anybody out there that has implemented or will be implementing the new DB2 MPP clustering on Linux?
http://www-3.ibm.com/software/data/db2/linux/ice/

Any success stories? Disasters?

Thanks in advance,

Tim

tim-at-datad-dot-com

You might want to check out the TPC-H benchmark document from IBM using the
ICE (or a similar) configuration:
http://www.tpc.org/tpch/results/tpch...p?id=103072902

Click on full disclosure report.
Nov 12 '05 #2
To which I reply, DUH!

I've ALREADY seen the stuff on the IBM website, what I want to know is if CUSTOMERS ( plural, more
than one ) have actually installed and are using this besides just the one example on ibm.com. In plain
english I want to hear from people that actually are doing this, not necessarily IBM. If someone from IBM
wants to tell me about a few customers that have actually installed the ICE system and are actually in
production then that's great too--but not the one they use in the ad. It's great to advertise something, it's
another thing for a potential customer to decide whether or not to buy it based solely on advertising. Should
we spend time on it without knowing about any reference accounts that have been either successful or
where the wheels came off and it was a complete flop? I don't think so, thusly, the request I originally
posted.

Thanks!

Tim

"Mark A" <ma@switchboard.net> wrote in message news:7e***************@news.uswest.net...
You might want to check out the TPC-H benchmark document from IBM using the
ICE (or a similar) configuration:
http://www.tpc.org/tpch/results/tpch...p?id=103072902

Click on full disclosure report.


Nov 12 '05 #3
> "Mark A" <ma@switchboard.net> wrote in message
news:7e***************@news.uswest.net...
You might want to check out the TPC-H benchmark document from IBM using the ICE (or a similar) configuration:
http://www.tpc.org/tpch/results/tpch...p?id=103072902

Click on full disclosure report.

"Tim Schaefer" <we*******@datad.com> wrote in message
news:5w*******************@fe03.atl2.webusenet.com ... To which I reply, DUH!

I've ALREADY seen the stuff on the IBM website, what I want to know is if CUSTOMERS ( plural, more than one ) have actually installed and are using this besides just the one example on ibm.com. In plain english I want to hear from people that actually are doing this, not necessarily IBM. If someone from IBM wants to tell me about a few customers that have actually installed the ICE system and are actually in production then that's great too--but not the one they use in the ad. It's great to advertise something, it's another thing for a potential customer to decide whether or not to buy it based solely on advertising. Should we spend time on it without knowing about any reference accounts that have been either successful or where the wheels came off and it was a complete flop? I don't think so, thusly, the request I originally posted.

Thanks!

Tim

There is no reason to act like a jerk.

The link I provided is not on the IBM website, it is on the TPC website. It
represents the results of a real world benchmark (independently monitored)
that can be measured against other hardware/databases using the same data
and queries. While it might not answer all your concerns, and it might not
even be the exact ICE product, it should give some indication of how well it
performs compared to other solutions (both in absolute terms and relative
cost terms). They also document exactly how they did it (including install
scripts and database parms used). You can't get any more real world than
that.

There is really nothing new in the ICE other than the blade hardware that
enables for fairly cheap clusters. DB2 has run in a clustering environment
for a long time with similar (if not identical configurations).
Nov 12 '05 #4
Mark,

Thanks much for your help and suggestions. Sorry if my post reads like a jerk,
I am typically to-the-point or long-winded, email is not my strong point in life.
At work I typically will get up from my desk and walk to another persons' desk
instead of using the phone or email. Email usually gets me in trouble, and
I'm terrible with a phone other than dialing a number and chatting. In your case
we'll have to settle for the worst of skillsets, my apologies.

I did peruse the TPC site a week or so ago, but it is my belief that there are lies
damn lies and statistics, thusly I'd rather hear from real people who have actually
implemented the ICE and can let me know in advance whether or not we would
be spending money foolishly or wisely. It is always easy to suggest somebody
else buy something but more difficult to absorb the loss if the decision is horribly
wrong. DB2 continues to gain ground with me intellectually however I need to
hear that others have found it as appealing as it looks in the brochure enough
to buy, and it works as advertised.

Thanks,

Tim

"Mark A" <ma@switchboard.net> wrote in message news:FU******************@news.uswest.net...
"Mark A" <ma@switchboard.net> wrote in message

news:7e***************@news.uswest.net...
You might want to check out the TPC-H benchmark document from IBM using the ICE (or a similar) configuration:
http://www.tpc.org/tpch/results/tpch...p?id=103072902

Click on full disclosure report.


"Tim Schaefer" <we*******@datad.com> wrote in message
news:5w*******************@fe03.atl2.webusenet.com ...
To which I reply, DUH!

I've ALREADY seen the stuff on the IBM website, what I want to know is if

CUSTOMERS ( plural, more
than one ) have actually installed and are using this besides just the one

example on ibm.com. In plain
english I want to hear from people that actually are doing this, not

necessarily IBM. If someone from IBM
wants to tell me about a few customers that have actually installed the

ICE system and are actually in
production then that's great too--but not the one they use in the ad.

It's great to advertise something, it's
another thing for a potential customer to decide whether or not to buy it

based solely on advertising. Should
we spend time on it without knowing about any reference accounts that have

been either successful or
where the wheels came off and it was a complete flop? I don't think so,

thusly, the request I originally
posted.

Thanks!

Tim

There is no reason to act like a jerk.

The link I provided is not on the IBM website, it is on the TPC website. It
represents the results of a real world benchmark (independently monitored)
that can be measured against other hardware/databases using the same data
and queries. While it might not answer all your concerns, and it might not
even be the exact ICE product, it should give some indication of how well it
performs compared to other solutions (both in absolute terms and relative
cost terms). They also document exactly how they did it (including install
scripts and database parms used). You can't get any more real world than
that.

There is really nothing new in the ICE other than the blade hardware that
enables for fairly cheap clusters. DB2 has run in a clustering environment
for a long time with similar (if not identical configurations).


Nov 12 '05 #5
"Tim Schaefer" <we*******@datad.com> wrote in message
news:Sr********************@fe12.atl2.webusenet.co m...
Mark,

Thanks much for your help and suggestions. Sorry if my post reads like a jerk, I am typically to-the-point or long-winded, email is not my strong point in life. At work I typically will get up from my desk and walk to another persons' desk instead of using the phone or email. Email usually gets me in trouble, and I'm terrible with a phone other than dialing a number and chatting. In your case we'll have to settle for the worst of skillsets, my apologies.

I did peruse the TPC site a week or so ago, but it is my belief that there are lies damn lies and statistics, thusly I'd rather hear from real people who have actually implemented the ICE and can let me know in advance whether or not we would
be spending money foolishly or wisely. It is always easy to suggest somebody else buy something but more difficult to absorb the loss if the decision is horribly wrong. DB2 continues to gain ground with me intellectually however I need to hear that others have found it as appealing as it looks in the brochure enough to buy, and it works as advertised.

Thanks,

Tim

Tim,

Thanks for your post.

I think you need to approach the ICE solution from two different
perspectives: software and hardware.

The DB2 software is ESE with partitioning option running on Linux (in this
case with support for 64 bit AMD processors). The multi-node configuration
of DB2 runs on other Linux boxes also, so the ICE offering is not
necessarily new from that standpoint. DB2 ESE (and especially its
predecessor DB2 V7 EEE) is running on many other platforms, including AIX
and Windows, and there should be many customer references available, even
some on Linux (not using the ICE hardware).

The hardware component of the ICE offering is really just a bunch of PC
(usually 2 processors each) running AMD 64 bit processors with some shared
disk components. This is really not a true "shared nothing" architecture
(despite what IBM says) since the disks are shared, and in most DB2
configurations (like the TPC benchmark configuration) there are two DB2
partitions per node, which is not share nothing since they share the OS
memory (however one could configure DB2 with one partition per node if you
really wanted to).

Since the days when Teradata invented shared nothing parallel database
processing, huge enhancements in disk technology and memory management has
lessened the need for an absolutely shared nothing configuration (although
to repeat--DB2 will support shared nothing if you want to set it up that way
from a hardware standpoint). DB2 does not care whether you have shared disks
between nodes, or multiple partitions per node (except from a performance
standpoint). All databases that support multi-node parallel configuration
support shared nothing hardware if set up that way, but even Teradata
abandoned the absolute share nothing hardware about 8 years ago. ICE is a
"shared less" architecture in a relatively affordable package (compared to
other similar configurations).

So maybe if you narrowed down your concern to the software part or the
hardware part, you might be able to better figure out how comfortable you
are with the solution. You could even get a couple of AMD 64 bit boxes and
create a 2 node system and try out DB2 ESE with partitioning option in a
proof of concept if you are really concerned about the ICE. From my
perspective, the ICE is not really anything all that new, it just some
existing software and new hardware that has been packaged as a very
affordable MPP system.
Nov 12 '05 #6
Well, here is one DB@ ICE customer that I could find on the internet
(still a bit too fresh to brag about)
http://biz.yahoo.com/iw/031106/059666.html

I'll try to lure out top DB2 ICE pinguin into the NG

Cheers
Serge

--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab

Nov 12 '05 #7
Thanks for the penguin :-)

1.) I suggest to check out the DB2 ICE web pages under ibm.com/db2/linux and
then our papers section as well. We have a nice technical backgrounder about
a finance customer we implemented with ICE up there. this is not the same as
under 2.)

2.) Serge mentioend already e-trade so no need for repeats

3.) Largest production ICE cluster is right now 32 nodes in North America
(attention: ICE cluster means all after ICE and specific ICE not just
generic DB2 cluster for Linux for genom scanning for example...t hese larger
ones were rolled before and I do not rebatch customers)

4.) HA etc - look intot he papers section ont he web - we have tons of stuff
up there.

Chers,
Boris
"Serge Rielau" <sr*****@ca.eye-bee-m.com> wrote in message
news:bp**********@hanover.torolab.ibm.com...
Well, here is one DB@ ICE customer that I could find on the internet
(still a bit too fresh to brag about)
http://biz.yahoo.com/iw/031106/059666.html

I'll try to lure out top DB2 ICE pinguin into the NG

Cheers
Serge

--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab

Nov 12 '05 #8
Thanks Boris,
What do I owe you? A two-four of fish-sticks?

Cheers
Serge

--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab

Nov 12 '05 #9
"Mark A" <ma@switchboard.net> writes:
This is really not a true "shared nothing" architecture (despite what
IBM says) since the disks are shared, and in most DB2 configurations
(like the TPC benchmark configuration) there are two DB2 partitions
per node, which is not share nothing since they share the OS memory
(however one could configure DB2 with one partition per node if you
really wanted to).
Important thing to note is that disks are not shared across physical
nodes, the architecture allows you to share disks between physical nodes
(or use SAN) but it doesn't need to be, thus contrasting it to
shared-disk environments such as Oracle RAC where a shared disk is a
necessity.
Since the days when Teradata invented shared nothing parallel database
processing, huge enhancements in disk technology and memory management has
lessened the need for an absolutely shared nothing configuration.


In fact Teradata is setup the same way as DB2. On their 2-way systems
they create multiple vprocs (virtual processors). The reason a dbms is
called shared-nothing is that any communication between vprocs for TD
and MLNs (multiple logical nodes, aka, logical data partitions) for db2
is through the communication layer provided in the dbms (done over the
Bynet for TD, and tcp/ip (or shared memory when possible) for db2 by the
fcm process).

Regards,
--
Haider
Nov 12 '05 #10
"> "Mark A" <ma@switchboard.net> writes:
This is really not a true "shared nothing" architecture (despite what
IBM says) since the disks are shared, and in most DB2 configurations
(like the TPC benchmark configuration) there are two DB2 partitions
per node, which is not share nothing since they share the OS memory
(however one could configure DB2 with one partition per node if you
really wanted to).


Important thing to note is that disks are not shared across physical
nodes, the architecture allows you to share disks between physical nodes
(or use SAN) but it doesn't need to be, thus contrasting it to
shared-disk environments such as Oracle RAC where a shared disk is a
necessity.

They don't share physical disks, but they share some components on the SAN.
Sharing a SAN across nodes is not really a concern due to the great
improvements in disk technology since a true share nothing architecture was
used with teh "old" Teradata.
Since the days when Teradata invented shared nothing parallel database
processing, huge enhancements in disk technology and memory management has lessened the need for an absolutely shared nothing configuration.


In fact Teradata is setup the same way as DB2. On their 2-way systems
they create multiple vprocs (virtual processors). The reason a dbms is
called shared-nothing is that any communication between vprocs for TD
and MLNs (multiple logical nodes, aka, logical data partitions) for db2
is through the communication layer provided in the dbms (done over the
Bynet for TD, and tcp/ip (or shared memory when possible) for db2 by the
fcm process).

Regards,
Haider


You are talking about the "New" Teradata platform, which is NOT a
share-nothing architecture. The original Teradata architecture (abandoned
about 6-8 years ago) was true share-nothing and did not have vprocs. Each
node was a single Intel 386 (and later 486) box with its own memory, disk,
etc. Each node had one database partition. The hardware and 16 bit OS for
each node was proprietary.

Teradata/NCR has since moved to virtual nodes on standard SMP machines
running UNIX. The IBM ICE solution is closer to the old share-nothing
architecture, however, it is still not a true share-nothing system if there
are two DB2 partitions per node and a SAN is used.
Nov 12 '05 #11

This thread has been closed and replies have been disabled. Please start a new discussion.

Similar topics

0
by: Tony Barker | last post by:
Helpful associates, We are going to acquire and new database server that is intel based and we want to have some failover capabilities. Our question now is should we go with Linux based cluster...
0
by: Leonid | last post by:
From: leonid_n@infogateonline.com (Leonid) Newsgroups: comp.databases.ms-sqlserver Subject: SQL Agent + Cluster NNTP-Posting-Host: 80.74.102.10 Message-ID:...
4
by: Ravikanth[MVP] | last post by:
Hi It is possible that IIS and SQL Server can reside on Seperate Machines and you can use Integrated Windows Authentication to connect. Ravikanth >-----Original Message-----
2
by: Bill Moran | last post by:
Boy, I'm just full of questions today ... Is there a way to provide different persistent settings for DateStyle for different databases within the same cluster? Simply put, I've got two...
1
by: abhinav | last post by:
Hi guys.I have read that one cannot perform true multithreading in python due to global interpreter lock mechanism.Suppose i have to implement a crawler on a say cluster system like clusterknoppix...
4
by: ThunderMusic | last post by:
Hi, We have many servers setup as a cluster. When one server crashes, another one take the relay... We want to know if it's possible (I suppose it is) to make a Windows service developed with .NET...
2
by: ruediger.papke | last post by:
Hello, we want to build up a DB2- switchover cluster on 2 solaris machines, where a db2-instance is switched from one server to another. we have 2 machines (a,b) with OS solaris 5.9. both...
3
by: Bruno LIVERNAIS | last post by:
Hi, We are currently installing a DB2 V9 ESE on a Linux server (RHEL4U4-x86_64). Installation runs successfully on each node. Database user environment is OK and the instance is well created. To...
0
by: he59 | last post by:
Hi, We are currently installing a DB2 V9 ESE on a Linux server (RHEL4U4-x86_64). Installation runs successfully on each node. Database user environment is OK and the instance is well created....
0
by: taylorcarr | last post by:
A Canon printer is a smart device known for being advanced, efficient, and reliable. It is designed for home, office, and hybrid workspace use and can also be used for a variety of purposes. However,...
0
by: Charles Arthur | last post by:
How do i turn on java script on a villaon, callus and itel keypad mobile phone
0
by: aa123db | last post by:
Variable and constants Use var or let for variables and const fror constants. Var foo ='bar'; Let foo ='bar';const baz ='bar'; Functions function $name$ ($parameters$) { } ...
0
by: ryjfgjl | last post by:
If we have dozens or hundreds of excel to import into the database, if we use the excel import function provided by database editors such as navicat, it will be extremely tedious and time-consuming...
0
BarryA
by: BarryA | last post by:
What are the essential steps and strategies outlined in the Data Structures and Algorithms (DSA) roadmap for aspiring data scientists? How can individuals effectively utilize this roadmap to progress...
1
by: nemocccc | last post by:
hello, everyone, I want to develop a software for my android phone for daily needs, any suggestions?
0
marktang
by: marktang | last post by:
ONU (Optical Network Unit) is one of the key components for providing high-speed Internet services. Its primary function is to act as an endpoint device located at the user's premises. However,...
0
by: Hystou | last post by:
Most computers default to English, but sometimes we require a different language, especially when relocating. Forgot to request a specific language before your computer shipped? No problem! You can...
0
jinu1996
by: jinu1996 | last post by:
In today's digital age, having a compelling online presence is paramount for businesses aiming to thrive in a competitive landscape. At the heart of this digital strategy lies an intricately woven...

By using Bytes.com and it's services, you agree to our Privacy Policy and Terms of Use.

To disable or enable advertisements and analytics tracking please visit the manage ads & tracking page.